Text to speech
English text to speech
Studio-recorded English voices, built for voice agents that have to answer while the caller is still on the line. Play them below and hear how they handle numbers, acronyms and brand names.
Listen
5 English voices
These are our studio samples: the same voices the API serves, not a marketing edit.
Noah
Native recording
Olivia
Native recording
Gustav
Native recording
Samantha
Native recording
Nick
Native recording
Details
What we tuned for this language
Built for interruption
English is where most voice agents start, and an agent lives or dies on the gap before it starts speaking. Kugel 3 streams audio as the text arrives, so the first syllable leaves the server long before the sentence is finished.
Acronyms, numbers and money
Order references, amounts and abbreviations are the sentences a demo never contains and production is full of. Our normalizer decides where to read letters individually and where to read a word, before synthesis.
Brand names, said your way
Every company has a handful of words that generic TTS mangles. Put them in the pronunciation dictionary once and every English voice on your account says them identically.
FAQ
English text to speech: common questions
How many English voices are on this page?
There are 5. Each one is a separate speaker rather than one voice with a pitch shift, and the same Kugel 3 model also speaks the other 25 languages we support.
Is it fast enough for a live voice agent?
Yes. The API streams over WebSocket and starts returning audio before the text is complete, which is what keeps a phone call from sounding like a walkie-talkie. It drops directly into Pipecat and LiveKit pipelines.
Can I use it for commercial products?
Yes. Paid plans include commercial use, and there is a free tier to test with first. Pricing is per generated second, with no per-voice licence.
Put this voice in your product
One API for streaming speech, a pronunciation dictionary for your own vocabulary, and hosting inside the EU or inside your own data centre.